BoCo EDA - Andy Pickering

Introduction

• This project is looking at patterns of homelessness-related arrests and possible relationship to weather in Boulder County.
• It was started by Sam Zhang, who gathered and merged the data: https://github.com/samzhang111/boco-jail-analysis.
• Data is in a Google drive folder shared by Sam, which I have synced locally on my machine.
• This analysis is in a github repository: https://github.com/andypicke/boco_homeless

Hypotheses To Explore

• How does weather affect daily bookings? weather: downtown-boulder-weather.csv. bookings: bookings-with-transient-status.csv (you’ll need to aggregate this one up to the daily level)
• What proportion of jail bookings for housed/unhoused folks are for homelessness related charges? (by department? by year?) we pulled out charges that came from a list of antihomelessness ordinances compiled by some folks at Denver University, and marked them in the bookings dataset. If any of them were true, we marked the antihomeless column as true.

Weather

• Add more details about where weather data comes from (on Sam’s github?)
• What is SNWD? Some kind of accumulated snowfall?
• What are WT01,WT02 etc.?
• SNOW is missing data from about 2011-204?

suppressPackageStartupMessages(library(dplyr))
library(readr)
suppressPackageStartupMessages(library(lubridate))
library(ggplot2)

Load weather data

wea <- read_csv('/Users/Andy/Google Drive/boco-jail/downtown-boulder-weather.csv',col_types = cols())
glimpse(wea)

Observations: 6,400
Variables: 12
$ DATE <int> 20000101, 20000102, 20000103, 20000104, 20000105, 20000106, 20000107, 20000...
$ PRCP <dbl> 0.00, 0.00, 0.08, 0.00, 0.00, 0.00, 0.00, 0.00, 0.00, 0.00, 0.00, 0.00, 0.0...
$ SNOW <dbl> 0.0, 0.0, 2.2, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ...
$ SNWD <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ TMAX <dbl> 54, 40, 36, 49, 47, 42, 47, 50, 45, 42, 61, 58, 41, 64, 63, 49, 64, 51, 56,...
$ TMIN <dbl> 29, 22, 19, 13, 26, 16, 19, 23, 29, 30, 17, 41, 25, 25, 34, 22, 30, 35, 32,...
$ WT01 <dbl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ WT03 <dbl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ WT04 <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ WT05 <dbl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ WT06 <dbl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ WT11 <dbl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...

summary(wea)

      DATE               PRCP              SNOW             SNWD                TMAX       
 Min.   :20000101   Min.   :0.00000   Min.   : 0.0000   Length:6400        Min.   :  6.00  
 1st Qu.:20040519   1st Qu.:0.00000   1st Qu.: 0.0000   Class :character   1st Qu.: 53.00  
 Median :20081004   Median :0.00000   Median : 0.0000   Mode  :character   Median : 67.00  
 Mean   :20083356   Mean   :0.05655   Mean   : 0.2372                      Mean   : 65.99  
 3rd Qu.:20130222   3rd Qu.:0.01000   3rd Qu.: 0.0000                      3rd Qu.: 81.00  
 Max.   :20170714   Max.   :9.08000   Max.   :16.7000                      Max.   :102.00  
                    NA's   :5         NA's   :1190                         NA's   :1       
      TMIN             WT01           WT03          WT04                WT05     
 Min.   :-17.00   Min.   :1      Min.   :1      Length:6400        Min.   :1     
 1st Qu.: 27.00   1st Qu.:1      1st Qu.:1      Class :character   1st Qu.:1     
 Median : 38.00   Median :1      Median :1      Mode  :character   Median :1     
 Mean   : 38.17   Mean   :1      Mean   :1                         Mean   :1     
 3rd Qu.: 51.00   3rd Qu.:1      3rd Qu.:1                         3rd Qu.:1     
 Max.   : 77.00   Max.   :1      Max.   :1                         Max.   :1     
 NA's   :3        NA's   :6304   NA's   :5623                      NA's   :6337  
      WT06           WT11     
 Min.   :1      Min.   :1     
 1st Qu.:1      1st Qu.:1     
 Median :1      Median :1     
 Mean   :1      Mean   :1     
 3rd Qu.:1      3rd Qu.:1     
 Max.   :1      Max.   :1     
 NA's   :6370   NA's   :6331  

names(wea) <- tolower(names(wea))
wea$date <- lubridate::ymd(wea$date)
wea <- wea %>% select( date,prcp,snow,tmax,tmin)
head(wea)

Plot temperature timeseries

wea %>%
        ggplot(aes(date,tmax))+
        geom_point() +
        ylab('Max Temp') +
        ggtitle('Downtown Boulder Weather')

Plot precipiation timeseries

• I assume in inches?

wea %>%
        ggplot(aes(date,prcp))+
        geom_point() +
        ylim(0,3)

Plot snow timeseries

• I assume this is daily snow fall in inches?
• Looks like we are missing a chunk of snowfall data from about 2011-2014 (pretty sure it snowed during those years :) )

wea %>%
        ggplot(aes(date,snow))+
        geom_point() +
        ylim(0,3)

Bookings

Some info on the bookings-with-transient-status.csv data: - fta is “Failure to Appear” - ftc is “Failure to Comply” - (Sam) These are bookings in Boulder County Jail only. - (Sam) Each row is an individual booking. - (Sam) boulder means the arrest was made by Boulder PD - (Sam) Column transient for whether someone was homeless or not. - (Sam) I created indicator variables for a variety of antihomeless charges (that a local law school came up with). antihomeless is true if any of the antihomeless charges existed in the booking. So one source of error could be that if someone was arrested for an antihomeless charge AND a more serious offense, I still mark it as “antihomeless”. - (Sam) I don’t remember making any_antihomeless – I suspect it’s identical to antihomeless and introduced by accident - (Andy) There are two Booking Time columns? I will use booking_time for now, as it seems better. - (Andy) What are locations? Is that where they were booked? - (Andy) Booked and booking_time appear to be duplicates?

Load bookings data

bk <- read_csv('/Users/Andy/Google Drive/boco-jail/bookings-with-transient-status.csv',col_types = cols())

number of columns of result is not a multiple of vector length (arg 1)46578 parsing failures.
row # A tibble: 5 x 5 col     row     col               expected actual expected   <int>   <chr>                  <chr>  <chr> actual 1  1537 Case No no trailing characters     .0 file 2  1539 Case No no trailing characters     .0 row 3  1540 Case No no trailing characters     .0 col 4  1541 Case No no trailing characters     .0 expected 5  1542 Case No no trailing characters     .0 actual # ... with 1 more variables: file <chr>
... ................. ... ............................................. ........ ............................................. ...... ............................................. .... ............................................. ... ............................................. ... ............................................. ........ ............................................. ...... .......................................
See problems(...) for more details.

glimpse(bk)

Observations: 167,633
Variables: 30
$ Name                 <chr> "HOOD,AARON JAY", "LAWYER,KENNETH A", "AGUILAR TORRES,MIGUE...
$ Booked               <dttm> 2000-01-01 02:12:00, 2000-01-01 04:01:00, 2000-01-01 03:11...
$ Location             <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ DOB                  <date> 1975-10-08, 1958-09-29, 1972-11-14, 1953-06-07, 1961-09-09...
$ Race                 <chr> "W", "W", "W", "W", "W", "W", "W", "W", "W", "W", "W", "W",...
$ Sex                  <chr> "M", "M", "M", "M", "M", "M", "M", "M", "M", "M", "F", "M",...
$ Case No              <int> 991126052, 991001313, 991126053, 991031932, 991102428, 9911...
$ Arresting Agency     <chr> "UNIVERSITY OF COLORADO", "JAIL MITTS ONLY", "UNIVERSITY OF...
$ Arrest Date          <date> 1999-12-31, 2000-01-16, 1999-12-31, 1999-12-31, 1999-12-31...
$ camping              <chr> "False", "False", "False", "False", "False", "False", "Fals...
$ fta                  <chr> "True", "False", "False", "False", "False", "False", "False...
$ ftc                  <chr> "False", "False", "False", "False", "False", "False", "Fals...
$ booking_time         <dttm> 2000-01-01 02:12:00, 2000-01-01 04:01:00, 2000-01-01 03:11...
$ boulder              <chr> "False", "False", "False", "False", "False", "False", "Fals...
$ urination            <chr> "False", "False", "False", "False", "False", "False", "Fals...
$ vehicle_as_residence <chr> "False", "False", "False", "False", "False", "False", "Fals...
$ public_obstruct      <chr> "False", "False", "False", "False", "False", "False", "Fals...
$ public_trespass      <chr> "False", "False", "False", "False", "False", "False", "Fals...
$ begging              <chr> "False", "False", "False", "False", "False", "False", "Fals...
$ antihomeless         <chr> "False", "False", "False", "False", "False", "False", "Fals...
$ smoking              <chr> "False", "False", "False", "False", "False", "False", "Fals...
$ any_antihomeless     <chr> "False", "False", "False", "False", "False", "False", "Fals...
$ Address              <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ City                 <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ State                <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ ZIP Code             <dbl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ Booking Date         <date> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA...
$ Booking Time         <dbl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ Facility             <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ transient            <chr> "False", "False", "False", "False", "False", "False", "Fals...

Let’s clean it up a little

# 'Booking Time' appears to be a duplicate/bad column
bk <- bk %>% select(-`Booking Time`)
# make columns names lowercase and remove spaces
names(bk) <- tolower(names(bk))
names(bk) <- gsub(' ','_',names(bk))
# some vars make more sense as factors
bk$arresting_agency <- as.factor(bk$arresting_agency)
bk$location <- as.factor(bk$location)
bk$race <- as.factor(bk$race)
# add wkday,month,year so we can aggregate by those
bk$wkday <- lubridate::wday(bk$booking_date,label=TRUE)
bk$month_ <- lubridate::month(bk$arrest_date,label=TRUE)
bk$year <- lubridate::year(bk$arrest_date)
# modern reporting seems to start in 2000 (yearly totals go from less 100 before 2000, to order of 10,000 starting w/ 2000). Keep only 2000 on here.
bk <- bk %>% filter(year>=2000)
bk$booked_date <- lubridate::date(bk$booked)
# a bunch of vars should be logical
to_log <- function(a_col){
        as.logical(a_col)
}
cols_to_log <- c('camping','boulder','urination','vehicle_as_residence','public_obstruct','public_trespass','begging','antihomeless','smoking','any_antihomeless','transient','fta','ftc')
bk[cols_to_log] <- lapply(bk[cols_to_log],to_log)
glimpse(bk)

Observations: 163,939
Variables: 33
$ name                 <chr> "LAWYER,KENNETH A", "COPELAND,MARK WILLIAM", "DITZEL,HOWARD...
$ booked               <dttm> 2000-01-01 04:01:00, 2000-01-01 03:39:00, 2000-01-01 06:00...
$ location             <fctr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA...
$ dob                  <date> 1958-09-29, 1953-01-27, 1948-01-25, 1977-03-15, 1980-06-22...
$ race                 <fctr> W, W, W, W, W, W, W, W, W, W, W, W, W, W, W, W, W, W, W, W...
$ sex                  <chr> "M", "M", "M", "F", "M", "M", "M", "F", "M", "M", "M", "M",...
$ case_no              <int> 991001313, 1089421, 1031978, 1103574, 5, 1101849, 1089420, ...
$ arresting_agency     <fctr> JAIL MITTS ONLY, LAFAYETTE PD, BOULDER PD, LONGMONT PD, BO...
$ arrest_date          <date> 2000-01-16, 2000-01-01, 2000-01-01, 2000-01-01, 2000-01-01...
$ camping              <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ fta                  <lgl> FALSE, FALSE, TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, FALS...
$ ftc                  <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ booking_time         <dttm> 2000-01-01 04:01:00, 2000-01-01 03:39:00, 2000-01-01 06:00...
$ boulder              <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ urination            <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ vehicle_as_residence <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ public_obstruct      <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ public_trespass      <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ begging              <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ antihomeless         <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ smoking              <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ any_antihomeless     <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ address              <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ city                 <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ state                <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ zip_code             <dbl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ booking_date         <date> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA...
$ facility             <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ transient            <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ wkday                <ord> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ month_               <ord> Jan, Jan, Jan, Jan, Jan, Jan, Jan, Jan, Jan, Jan, Jan, Jan,...
$ year                 <dbl> 2000, 2000, 2000, 2000, 2000, 2000, 2000, 2000, 2000, 2000,...
$ booked_date          <date> 2000-01-01, 2000-01-01, 2000-01-01, 2000-01-01, 2000-01-01...

Distribution of ages

• some bad data points? (ie age=-1 or age=133)

summary(ages)

   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
   -1.0    22.0    29.0    31.9    40.0   133.0 

What percent of arrestees are transient?

mean(bk$transient,na.rm = TRUE)*100

[1] 13.05669

What percent of arrests are ‘antihomeless’ (true if any charges were related to ‘antihomeless charges’)

mean(bk$antihomeless)*100

[1] 1.317563

What percent of arrests were made by Boulder PD?

mean(bk$boulder)*100

[1] 6.589646

levels(bk$arresting_agency)

 [1] "BOULDER COUNTY DRUG TASK FORCE" "BOULDER COUNTY SHERIFFS OFFICE"
 [3] "BOULDER PD"                     "COLORADO STATE PATROL"         
 [5] "COMMUNITY CORRECTIONS"          "DISTRICT ATTORNEYS OFFICE"     
 [7] "ERIE PD"                        "JAIL MITTS ONLY"               
 [9] "LAFAYETTE PD"                   "LONGMONT PD"                   
[11] "LOUISVILLE PD"                  "NEDERLAND MARSHALS OFFICE"     
[13] "OTHER"                          "PAROLE"                        
[15] "STATE DIVISION OF WILDLIFE"     "UNIVERSITY OF COLORADO"        
[17] "WARD MARSHALS OFFICE"          

What agencies have the most arrests?

bk %>%
        group_by(arresting_agency) %>%
        tally() %>%
        arrange( desc(n) ) %>%
        ggplot(aes(x=reorder(arresting_agency,n),y=n))+
        geom_bar(stat='identity',aes(fill=arresting_agency)) +
        coord_flip()

Number of arrests by race

bk %>% group_by(race) %>% 
        tally() %>% 
        arrange(desc(n)) %>% 
        ggplot(aes(x=reorder(race,n),y=n))+
        geom_bar(stat='identity',aes(fill=race)) +
        coord_flip()

Aggregate Monthly

• I thought there would be more arrests in winter, but there doesn’t seem to be that big of a difference.

bk %>% 
        filter(year %in% c(2011,2012,2013,2015,2015)) %>%
        group_by(year,month_) %>%
        tally() %>%
        ggplot(aes(month_,n))+
        geom_point() +
        geom_bar(stat='identity',aes(fill=month_)) +
        facet_wrap(~year)

Aggregate by wkday

bk %>%
        filter(!is.na(wkday)) %>%
        group_by(wkday) %>%
        tally() %>%
        ggplot( aes(wkday,n)) +
        geom_col(aes(fill=wkday)) +
        ggtitle('Total Arrests By Day, for ALL data')

By weekday, for separate years

bk %>%
        filter(!is.na(wkday)) %>%
        filter(year>1999) %>%
        group_by(year,wkday) %>%
        tally() %>%
        ggplot( aes(wkday,n)) +
        geom_col(aes(fill=wkday)) +
        facet_wrap(~year) +
        ggtitle('Total Arrests By Day, for each year')

By weekday, for separate months

bk %>%
        filter(!is.na(wkday)) %>%
        filter(year>1999) %>%
        group_by(month_,wkday) %>%
        tally() %>%
        ggplot( aes(wkday,n)) +
        geom_col(aes(fill=wkday)) +
        facet_wrap(~month_) +
        ggtitle('Total Arrests By Day, for each month, includes all years')

Try aggregating to daily level?

• Looks like daily number of arrests is decreasing over time?
• Not sure if actual, or something to do with how data is recorded.

bk %>% 
        filter(arrest_date>"2000-01-01") %>%
        group_by(arrest_date) %>%
        tally() %>%
        ggplot(aes(arrest_date,n)) +
        geom_point(alpha=0.2) +
        ylim(0,60) +
        geom_smooth(method="lm")

NA

Relationship between arrests and weather

Let’s merge the booking and weather data first

bk_wea <- left_join(bk,wea,by=c('arrest_date'='date'))
glimpse(bk_wea)

Observations: 163,939
Variables: 37
$ name                 <chr> "LAWYER,KENNETH A", "COPELAND,MARK WILLIAM", "DITZEL,HOWARD...
$ booked               <dttm> 2000-01-01 04:01:00, 2000-01-01 03:39:00, 2000-01-01 06:00...
$ location             <fctr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA...
$ dob                  <date> 1958-09-29, 1953-01-27, 1948-01-25, 1977-03-15, 1980-06-22...
$ race                 <chr> "W", "W", "W", "W", "W", "W", "W", "W", "W", "W", "W", "W",...
$ sex                  <chr> "M", "M", "M", "F", "M", "M", "M", "F", "M", "M", "M", "M",...
$ case_no              <int> 991001313, 1089421, 1031978, 1103574, 5, 1101849, 1089420, ...
$ arresting_agency     <fctr> JAIL MITTS ONLY, LAFAYETTE PD, BOULDER PD, LONGMONT PD, BO...
$ arrest_date          <date> 2000-01-16, 2000-01-01, 2000-01-01, 2000-01-01, 2000-01-01...
$ camping              <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ fta                  <lgl> FALSE, FALSE, TRUE, FALSE, FALSE, FALSE, FALSE, FALSE, FALS...
$ ftc                  <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ booking_time         <dttm> 2000-01-01 04:01:00, 2000-01-01 03:39:00, 2000-01-01 06:00...
$ boulder              <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ urination            <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ vehicle_as_residence <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ public_obstruct      <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ public_trespass      <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ begging              <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ antihomeless         <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ smoking              <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ any_antihomeless     <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ address              <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ city                 <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ state                <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ zip_code             <dbl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ booking_date         <date> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA...
$ facility             <chr> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ transient            <lgl> FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FALSE, FAL...
$ wkday                <ord> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,...
$ month_               <ord> Jan, Jan, Jan, Jan, Jan, Jan, Jan, Jan, Jan, Jan, Jan, Jan,...
$ year                 <dbl> 2000, 2000, 2000, 2000, 2000, 2000, 2000, 2000, 2000, 2000,...
$ booked_date          <date> 2000-01-01, 2000-01-01, 2000-01-01, 2000-01-01, 2000-01-01...
$ prcp                 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
$ snow                 <dbl> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,...
$ tmax                 <dbl> 49, 54, 54, 54, 54, 54, 54, 51, 54, 54, 54, 49, 54, 54, 54,...
$ tmin                 <dbl> 22, 29, 29, 29, 29, 29, 29, 35, 29, 29, 29, 22, 29, 29, 29,...

Scatter plot daily arrests vs weather variables

• There doesn’t seem to be much of a relationship between daily # arrests and temperature.

Scatter plot daily transient arrests vs weather variables

• Not too much for transient only either.

## Scatter plot daily antihomeless arrests vs weather variables

bk%>%
        filter(antihomeless==TRUE) %>% 
        group_by(arrest_date) %>% 
        tally() %>% 
        left_join(wea,by=c('arrest_date'='date')) %>%
        ggplot(aes(tmin,n)) +
        geom_point() +
        geom_smooth(method = 'lm')

Try a logistic regression of antihomeless arrest as function of temperature, precip, etc.?

Look at individual years?

bk_wea %>%
        filter(year==2002) %>% 
        group_by(arrest_date) %>% 
        tally() %>% 
        left_join(wea,by=c('arrest_date'='date')) %>%
        ggplot(aes(tmin,n)) +
        geom_point() +
        geom_smooth(method = 'lm')

bk %>% 
        group_by(arrest_date) %>% 
        tally() %>% 
        left_join(wea,by=c('arrest_date'='date')) %>% 
        ggplot(aes(arrest_date,tmin)) +
        geom_point() +
        geom_point(aes(x=arrest_date,y=n),color='red')